[2f5ca80] | 1 | #!/usr/bin/env python |
---|
| 2 | # -*- coding: utf-8 -*- |
---|
| 3 | """Comparison of the created MARS requests of two flex_extract versions. |
---|
| 4 | |
---|
| 5 | There will be comparisons for the given standard control files in the |
---|
| 6 | "Controls" - directory. The result of the comparison is stored in the |
---|
| 7 | "Log" - directory with an individual timestamp in the format %Y-%m-%d_%H-%M-%S. |
---|
| 8 | (E.g. log_2018-11-23_12-42-29) |
---|
| 9 | The MARS request files are named such that it contains the name of the |
---|
| 10 | corresponding control files "<control-identifier>.csv" (e.g. EA5_mr.csv). |
---|
| 11 | They are stored in the corresponding version directory and have the same |
---|
| 12 | name in both versions. |
---|
| 13 | |
---|
| 14 | The script should be called like: |
---|
| 15 | |
---|
| 16 | python test_cmp_mars_requests.py <old_version_number> <new_version_number> |
---|
| 17 | |
---|
| 18 | Note |
---|
| 19 | ---- |
---|
| 20 | The MARS request files from the older version have to be in place already. |
---|
| 21 | The request files of the new/current version will be generated automatically |
---|
| 22 | with the "run_local.sh" script. |
---|
| 23 | It is necessary to have a directory named after the version number of |
---|
| 24 | flex_extract. For example: "7.0.3" and "7.1". |
---|
| 25 | |
---|
| 26 | Example |
---|
| 27 | ------- |
---|
[6275f7b] | 28 | python test_cmp_mars_requests.py 7.0.4 7.1 |
---|
[2f5ca80] | 29 | """ |
---|
| 30 | |
---|
| 31 | # ------------------------------------------------------------------------------ |
---|
| 32 | # MODULES |
---|
| 33 | # ------------------------------------------------------------------------------ |
---|
| 34 | import os |
---|
| 35 | import sys |
---|
| 36 | import pandas as pd |
---|
| 37 | import subprocess |
---|
| 38 | import shutil |
---|
| 39 | from datetime import datetime |
---|
| 40 | |
---|
| 41 | sys.path.append('../../../source/python') |
---|
| 42 | import _config |
---|
[6275f7b] | 43 | from mods.tools import init128 |
---|
[2f5ca80] | 44 | |
---|
| 45 | # ------------------------------------------------------------------------------ |
---|
| 46 | # FUNCTION |
---|
| 47 | # ------------------------------------------------------------------------------ |
---|
| 48 | def test_mr_column_equality(mr_old, mr_new): |
---|
| 49 | '''Check if old and new version of MARS requests have the same |
---|
| 50 | amount of columns. |
---|
| 51 | |
---|
| 52 | If the number is not equal and/or the column names are not equal |
---|
| 53 | an error message is stored in global variable "err_msg". |
---|
| 54 | |
---|
| 55 | Parameters |
---|
| 56 | ---------- |
---|
| 57 | mr_old : :obj:`pandas DataFrame` |
---|
| 58 | Contains the mars requests from the old version of |
---|
| 59 | flex_extract. |
---|
| 60 | |
---|
| 61 | mr_new : :obj:`pandas DataFrame` |
---|
| 62 | Contains the mars requests from the new version of |
---|
| 63 | flex_extract. |
---|
| 64 | |
---|
| 65 | Return |
---|
| 66 | ------ |
---|
| 67 | bool |
---|
| 68 | True if successful, False otherwise. |
---|
| 69 | ''' |
---|
| 70 | global err_msg |
---|
| 71 | if (len(mr_old.columns.values) == len(mr_new.columns.values) and |
---|
| 72 | sorted(mr_old.columns.values) == sorted(mr_new.columns.values)): |
---|
| 73 | return True |
---|
| 74 | else: |
---|
| 75 | err_msg = 'Unequal number and/or column names!\n' |
---|
| 76 | return False |
---|
| 77 | |
---|
| 78 | |
---|
| 79 | def test_mr_number_equality(mr_old, mr_new): |
---|
| 80 | '''Check if old and new version have the same number of requests. |
---|
| 81 | |
---|
| 82 | If the number is not equal an error message is stored in |
---|
| 83 | global variable "err_msg". |
---|
| 84 | |
---|
| 85 | Parameters |
---|
| 86 | ---------- |
---|
| 87 | mr_old : :obj:`pandas DataFrame` |
---|
| 88 | Contains the mars requests from the old version of |
---|
| 89 | flex_extract. |
---|
| 90 | |
---|
| 91 | mr_new : :obj:`pandas DataFrame` |
---|
| 92 | Contains the mars requests from the new version of |
---|
| 93 | flex_extract. |
---|
| 94 | |
---|
| 95 | Return |
---|
| 96 | ------ |
---|
| 97 | bool |
---|
| 98 | True if successful, False otherwise. |
---|
| 99 | ''' |
---|
| 100 | global err_msg |
---|
| 101 | if len(mr_new.index) == len(mr_old.index): |
---|
| 102 | return True |
---|
| 103 | else: |
---|
| 104 | err_msg = 'Unequal number of mars requests!\n' |
---|
| 105 | return False |
---|
| 106 | |
---|
| 107 | def test_mr_content_equality(mr_old, mr_new): |
---|
| 108 | '''Check if old and new version have the same request contents. |
---|
| 109 | |
---|
| 110 | If the content in a column is not equal an error message is stored in |
---|
| 111 | global variable "err_msg", recording the column. |
---|
| 112 | |
---|
| 113 | Parameters |
---|
| 114 | ---------- |
---|
| 115 | mr_old : :obj:`pandas DataFrame` |
---|
| 116 | Contains the mars requests from the old version of |
---|
| 117 | flex_extract. |
---|
| 118 | |
---|
| 119 | mr_new : :obj:`pandas DataFrame` |
---|
| 120 | Contains the mars requests from the new version of |
---|
| 121 | flex_extract. |
---|
| 122 | |
---|
| 123 | Return |
---|
| 124 | ------ |
---|
| 125 | bool |
---|
| 126 | True if successful, False otherwise. |
---|
| 127 | ''' |
---|
| 128 | global err_msg |
---|
| 129 | lresult = None |
---|
| 130 | columns = list(mr_new.columns.values) |
---|
| 131 | del columns[columns.index('target')] |
---|
[6275f7b] | 132 | mr_new = trim_all_columns(mr_new) |
---|
| 133 | mr_old = trim_all_columns(mr_old) |
---|
[2f5ca80] | 134 | for col in columns: |
---|
| 135 | if mr_new[col].equals(mr_old[col]): |
---|
| 136 | lresult = True |
---|
| 137 | else: |
---|
| 138 | err_msg += 'Unconsistency happend to be in column: ' + col + '\n' |
---|
[6275f7b] | 139 | print mr_new[col] |
---|
| 140 | print mr_old[col] |
---|
[2f5ca80] | 141 | return False |
---|
| 142 | return lresult |
---|
| 143 | |
---|
| 144 | |
---|
[6275f7b] | 145 | def trim_all_columns(df): |
---|
| 146 | """ |
---|
| 147 | Trim whitespace from ends of each value across all series in dataframe |
---|
| 148 | """ |
---|
| 149 | trim_strings = lambda x: x.strip() if isinstance(x, str) else x |
---|
| 150 | return df.applymap(trim_strings) |
---|
| 151 | |
---|
| 152 | def convert_param_numbers(mr_old): |
---|
| 153 | """ |
---|
| 154 | Convert the numbers parameter into integers with 3 digits. |
---|
| 155 | """ |
---|
| 156 | |
---|
| 157 | if str(mr_old).strip() != "OFF" and mr_old != None and '/' in str(mr_old) : |
---|
| 158 | numbers = mr_old.split('/') |
---|
| 159 | number = str(int(numbers[0])).zfill(3)+'/TO/'+str(int(numbers[2])).zfill(3) |
---|
| 160 | |
---|
| 161 | return number |
---|
| 162 | |
---|
| 163 | return mr_old |
---|
| 164 | |
---|
| 165 | def convert_param_step(mr_old): |
---|
| 166 | """ |
---|
| 167 | For pure forecast with steps greater than 23 hours, the older versions |
---|
| 168 | writes out a list of steps instead with the syntax 'to' and 'by'. |
---|
| 169 | e.g. 000/003/006/009/012/015/018/021/024/027/030/033/036 |
---|
| 170 | |
---|
| 171 | Convert this to 0/to/36/by/3 |
---|
| 172 | """ |
---|
| 173 | |
---|
| 174 | #if 'to' in str(mr_old) and 'by' in str(mr_old): |
---|
| 175 | # steps = mr_old.split('/') |
---|
| 176 | # step = [] |
---|
| 177 | # for i in range(int(steps[0]),int(steps[2]),int(steps[4])): |
---|
| 178 | # step.append(str(int(i)).zfill(2)) |
---|
| 179 | # return '/'.join(step) |
---|
| 180 | |
---|
| 181 | if not mr_old.isdigit() and 'to' not in mr_old.lower(): |
---|
| 182 | if int(mr_old.split('/')[-1]) > 23: |
---|
| 183 | |
---|
| 184 | steps = mr_old.split('/') |
---|
| 185 | dtime = int(steps[1]) - int(steps[0]) |
---|
| 186 | |
---|
| 187 | nsteps = str(int(steps[1]))+'/to/'+str(int(steps[-1]))+'/by/'+str(int(dtime)) |
---|
| 188 | return nsteps |
---|
| 189 | |
---|
| 190 | |
---|
| 191 | return mr_old |
---|
| 192 | |
---|
| 193 | def to_param_id(pars, table): |
---|
| 194 | '''Transform parameter names to parameter ids with ECMWF grib table 128. |
---|
| 195 | |
---|
| 196 | Parameters |
---|
| 197 | ---------- |
---|
| 198 | pars : str |
---|
| 199 | Addpar argument from CONTROL file in the format of |
---|
| 200 | parameter names instead of ids. The parameter short |
---|
| 201 | names are sepearted with "/" and they are passed as |
---|
| 202 | one single string. |
---|
| 203 | |
---|
| 204 | table : dict |
---|
| 205 | Contains the ECMWF grib table 128 information. |
---|
| 206 | The key is the parameter number and the value is the |
---|
| 207 | short name of the parameter. |
---|
| 208 | |
---|
| 209 | Return |
---|
| 210 | ------ |
---|
| 211 | ipar : list of int |
---|
| 212 | List of addpar parameters from CONTROL file transformed to |
---|
| 213 | parameter ids in the format of integer. |
---|
| 214 | ''' |
---|
| 215 | if not pars: |
---|
| 216 | return [] |
---|
| 217 | if not isinstance(pars, str): |
---|
| 218 | pars=str(pars) |
---|
| 219 | |
---|
| 220 | cpar = pars.upper().split('/') |
---|
| 221 | spar = [] |
---|
| 222 | for par in cpar: |
---|
| 223 | par = par.strip() |
---|
| 224 | for k, v in table.items(): |
---|
| 225 | if par.isdigit(): |
---|
| 226 | par = str(int(par)).zfill(3) |
---|
| 227 | if par == k or par == v: |
---|
| 228 | spar.append(k + '.128') |
---|
| 229 | break |
---|
| 230 | else: |
---|
| 231 | print('\n\Warning: par ' + par + ' not found in table 128\n\n”') |
---|
| 232 | |
---|
| 233 | return '/'.join(str(l) for l in spar) |
---|
| 234 | |
---|
| 235 | |
---|
| 236 | |
---|
[2f5ca80] | 237 | if __name__ == '__main__': |
---|
| 238 | |
---|
| 239 | # basic values for paths and versions |
---|
| 240 | control_path = 'Controls' |
---|
| 241 | log_path = 'Log' |
---|
[6275f7b] | 242 | old_dir = sys.argv[1] # e.g. '7.0.4' |
---|
[2f5ca80] | 243 | new_dir = sys.argv[2] # e.g. '7.1' |
---|
| 244 | mr_filename = 'mars_requests.csv' |
---|
| 245 | |
---|
| 246 | # have to be set to "True" in the beginnning |
---|
| 247 | # so it only fails if a test fails |
---|
| 248 | lfinal = True |
---|
| 249 | |
---|
| 250 | # prepare log file for this test run |
---|
| 251 | currenttime = datetime.now() |
---|
| 252 | time_str = currenttime.strftime('%Y-%m-%d_%H-%M-%S') |
---|
| 253 | logfile = os.path.join(log_path, 'log_' + time_str) |
---|
| 254 | with open(logfile, 'aw') as f: |
---|
| 255 | f.write('Compare mars requests between version ' + old_dir + |
---|
| 256 | ' and version ' + new_dir + ' : \n') |
---|
| 257 | |
---|
| 258 | # list all controlfiles |
---|
| 259 | controls = os.listdir(control_path) |
---|
| 260 | |
---|
| 261 | # loop over controlfiles |
---|
| 262 | for c in controls: |
---|
| 263 | # empty error message for every controlfile |
---|
| 264 | err_msg = '' |
---|
| 265 | |
---|
| 266 | # start flex_extract with controlfiles to get mars_request files |
---|
| 267 | shutil.copy(os.path.join(control_path,c), _config.PATH_CONTROLFILES) |
---|
| 268 | subprocess.check_output(['run_local.sh', new_dir, c]) |
---|
| 269 | os.remove(os.path.join(_config.PATH_CONTROLFILES,c)) |
---|
| 270 | |
---|
| 271 | # cut-of "CONTROL_" string and mv mars_reqeust file |
---|
| 272 | # to a name including control specific name |
---|
| 273 | mr_name = c.split('CONTROL_')[1] + '.csv' |
---|
| 274 | shutil.move(os.path.join(new_dir,mr_filename), os.path.join(new_dir,mr_name)) |
---|
| 275 | |
---|
| 276 | # read both mr files (old & new) |
---|
| 277 | mr_old = pd.read_csv(os.path.join(old_dir, mr_name)) |
---|
| 278 | mr_new = pd.read_csv(os.path.join(new_dir, mr_name)) |
---|
| 279 | |
---|
| 280 | mr_old.columns = mr_old.columns.str.strip() |
---|
| 281 | mr_new.columns = mr_new.columns.str.strip() |
---|
| 282 | |
---|
[6275f7b] | 283 | # convert names in old to ids |
---|
| 284 | table128 = init128(_config.PATH_GRIBTABLE) |
---|
| 285 | #print mr_old['param'] |
---|
| 286 | |
---|
| 287 | # some format corrections are necessary to compare older versions with 7.1 |
---|
| 288 | mr_old['param'] = mr_old['param'].apply(to_param_id, args=[table128]) |
---|
| 289 | mr_old['number'] = mr_old['number'].apply(convert_param_numbers) |
---|
| 290 | if '142' in mr_old.ix[0,'param']: # if flux request |
---|
| 291 | mr_old.ix[0,'step'] = convert_param_step(mr_old.ix[0,'step']) |
---|
| 292 | |
---|
| 293 | print 'Results: ', c |
---|
| 294 | |
---|
[2f5ca80] | 295 | # do tests on mr files |
---|
| 296 | lcoleq = test_mr_column_equality(mr_old, mr_new) |
---|
| 297 | lnoeq = test_mr_number_equality(mr_old, mr_new) |
---|
| 298 | lcoeq = test_mr_content_equality(mr_old, mr_new) |
---|
| 299 | |
---|
| 300 | # check results for mr file |
---|
| 301 | lfinal = lfinal and lcoleq and lnoeq and lcoeq |
---|
| 302 | |
---|
| 303 | # write out result to logging file |
---|
| 304 | with open(logfile, 'aw') as f: |
---|
| 305 | if lcoleq and lnoeq and lcoeq: |
---|
| 306 | f.write('... ' + c + ' ... OK!' + '\n') |
---|
| 307 | else: |
---|
| 308 | f.write('... ' + c + ' ... FAILED!' + '\n') |
---|
| 309 | if err_msg: |
---|
| 310 | f.write('... \t' + err_msg + '\n') |
---|
| 311 | |
---|
[6275f7b] | 312 | exit |
---|
| 313 | |
---|
[2f5ca80] | 314 | # exit with success or error status |
---|
| 315 | if lfinal: |
---|
| 316 | sys.exit(0) # 'SUCCESS' |
---|
| 317 | else: |
---|
| 318 | sys.exit(1) # 'FAIL' |
---|